Download Analysis compliance using claims database

The Impact of Patient Adherence on
Physician Performance Measurement
Michael B. Nichol, Ph.D.
Pharmaceutical Economics and Policy
University of Southern California
1
Presentation Objectives




The nature of the problem: Why do adherence and
persistence impact P4P?
The problem metric: How do we measure adherence
and persistence?
The population: Who adheres or persists?
Solving the problem: What can we do to improve our
performance, especially for P4P?
2
Definitions




Why are we going to use compliance and adherence
interchangeably?
Compliance = whether patients follow the instructions of
their doctor
 Dichotomous measure
Adherence = whether patients endorse the instructions
of their doctor
 Dichotomous measure
Persistence = how long they follow the advice (whether
they modify it over time)
 dichotomous or continuous
3
Motivation for Compliance Studies



General recognition that non-compliance is a problem
Ultimate goal is to improve health outcome by targeting
some patients on modifiable factors to improve
compliance
Different parties in health care have different
perspectives and interest (e.g., clinicians, patients, and
payers)
4
Evidence that Non-Compliance is a Problem:
Medication Event Monitoring System (MEMS)
5
Administrative Data
on Drug Reimbursement
Information about patient’s medication acquisition and
procurement behavior using pharmaceutical benefit manager
(PBM) reimbursement data
Features
 Good snapshot of acquisition
in relationship to other
mediations
 Affordable
 Reliability can be ascertained
 Insight into dosing intervals
Limitations
 Many confounders
 Selection bias
 Dependent on
provider of data
 Retrospective
6
Advantages of Administrative
Claims-based Analyses







Objective (no recall bias)
Real-world (not controlled)
Relatively cheap to obtain
Large sample
Multiple outcomes
Cost analysis
Pattern recognition
7
Measuring the Complexity
of Non-Compliance
Drug A
Drug B
8
An Example of Non-Compliance: Statins
Benner et al. JAMA 2002;288:455-461.
9
Days Covered
for Statins and CCBs
100
Proportion of patient in cohort (%)
90
80
70
Statin only
Statin+CCB*
CCB only
CCB+Statin*
Statin&CCB
60
50
40
30
20
10
0
1
31
61
91
121
151
181
211
241
271
301
331
361
Day
Source: Unpublished data
10
Medication Possession Ratio for Statins/CCBs
Figure 1. Average 180-Day MPR by Index Year
Average 180-Day MPR (%)
80
CCB only
Statin only
Statin+CCB
Statin*+CCB
CCB+Statin
CCB*+Statin
70
60
* MPR is for the
newly initiated
(not current)
medication in the
combination
therapy.
50
40
30
1995
1996
1997
1998
1999
2000
2001
Index Year
Source: Unpublished data
11
Patterns of Non-Compliance

15-20% of first scripts never filled

Of those filled, 20-35% never fill a second

Persistence declines slowly after 6 months

Patients who discontinue rarely restart, at least
within a two year window for many chronic
problems
12
Why Do We Care?




Lack of efficacy from recommended treatment
Increased mortality and morbidity
Increased costs
Inability to meet P4P goals
13
Health Care Costs Associated with Discontinuation:
Hypertension
$600
$281
$400
$200
$0
($200)
($400)
($600)
($800)
($174)
($591)
($637)
Type of Service
Ambulatory
Hospital
Drugs
Total Costs
N=6,430
Source: McCombs JS, Nichol MB, Newman C and Sclar DA: The costs of interrupting
antihypertensive drug therapy in a Medicaid population. Medical Care, 32(3): 214-226, 1994.
14
Health Care Costs Associated with Discontinuation:
Major Depressive Disorder
$600
$515
$100
($157)
($400)
($900)
($753)
($922)
($1,400)
Type of Service
Ambulatory
Hospital
Drugs
Total Costs
N=1,240
Source: McCombs JS, Nichol MB, Stimmel GL. The role of SSRI antidepressants for treating depressed
15
patients in the California Medicaid (Medi-Cal) program. Value in Health, 2(4): 269-280, 1999.
Goal Attainment
Efficacy/Goal Relationship
100%
100%
Efficacy
16
Goal Attainment
Efficacy/Goal/Adherence
Relationship
100%
10% Efficacy reduction
Plus
30% Adherence reduction
100%
Efficacy
17
Impact on P4P Goals:
LDL <= 130





Simulation of impact on P4P LDL goal
Data source is the IHA 2007 P4P reporting for
LDL <= 130
84.3% of the cardiovascular population were
screened
Assumes that all patients screened are treated
Assumes that treatment is 100% efficacious
18
Impact on P4P Goals:
LDL <= 130
Proportion
adherent to Rx
baseline
All screened
90% screened
treated, Proportion treated, Proportion
meeting goal
meeting goal
59.6%
59.6%
(70% “adherent”) (78% “adherent”)
80% Rx adherent
67.4%
60.7%
90% Rx adherent
75.9%
68.3%
Source: Table computed from 2007 P4P results provided at iha.org
19
Theory of Compliance Behavior
20
Multiple Causes
(current or recent)
Factor level
Ease of tak i n g, dosi n g fre qu e n cy, taste ,
Bran d l oyal ty, S i de e ffe cts
Latent Construct
Medication
(treatment)
Level
Di se ase S tatu s
S ati sfacti on
Generalized Concept of Compliance
Be l i e fs
S e l f-e ffi cacy
Individual
Level
Compliance
Pe rson al i ty (forge tfu l n e ss,
re spon si ve n e ss)
Locu s of con trol
Un i de n ti fi e d
Patient-Physician Relations (ex pert power)
Interpersonal
Level
Fami l y an d S oci al S u pport (re fe re n t
powe r)
In su ran ce (copay)
Pol i cy (re sou rce , care acce ss)
C u l tu ral n orm S oci al S tan dard
Community
Level
21
Target of compliancepromoting intervention
Causes of Non-Compliance



Multiple causes with multiple levels
Many factors may not be observable to researchers
(many latent variables)
Each causal level can be targeted to improve
compliance
22
What Can We Do About Non-Adherence:
Targeting for Compliance




Who is non-compliant?
Why are they non-compliant?
How can we change their behavior?
When can we change their behavior?
23
Demographic Associations with Adherence


Few studies show clear correlations with adherence
among characteristics like age, gender, education, and
socio-economic status
Correlation between patient education level and
adherence is positive, but only for medications to treat
chronic disease
DiMatteo MR. Variations in patients’ adherence to medical recommendations,
Medical Care: 42:200-209, 2004
24
Selected Disease Condition
Adherence Rates
Mean Rate
100
90
80
Mean Rate
70
60
50
40
30
20
10
0
HIV (N=8)
Cancer (N=65)
CVD (N=129)
Infectious disease
(N=34)
Diabetes (N=23)
DiMatteo MR. Variations in patients’ adherence to medical recommendations, Medical Care: 42:200-209, 2004
25
Reasons for Non-Compliance
Don’t like being told
what to do (0.6%)
Too expensive (1.8%)
Don’t like being dependent
on drugs (7.3%)
Other (3.6%)
I just forget (54.9%)
If I don’t take them, supply
will last longer (1.3%)
Side effects (6.4%)
Don’t think drugs are
working (3.4%)
Hate taking drugs (7.1%)
Don’t think its always
necessary (13.7%)
Cheng JW, et al. Pharmacotherapy. 2001;21:828-841.
26
What Works? A Review of Reviews


Review of 38 meta-analytic reviews of adherence
interventions
Technical Interventions (simplifying medication regimen;
electronic monitoring)
 Less frequent dosing = improved adherence
 Single dose/day better than multiple doses/day
 Electronic device improvements attributed to
reduction in doses
Van Dulmen S, et al. Patient adherence to medical treatment: A review of reviews, BMC Health
Services Research, 7:55, 2007
27
What Works? A Review of Reviews

Behavioral Interventions (memory aids, monitoring by
calendars, support or rewards)
 Financial rewards improved adherence in 10/11
studies
 Mail and telephone reminders can improve adherence
Van Dulmen S, et al. Patient adherence to medical treatment: A review of reviews, BMC Health
Services Research, 7:55, 2007
28
What Works? A Review of Reviews

Educational Interventions (teaching/providing
knowledge, including personal, group, audio-visual,
home visits)
 Effects can be important in the short term, but decay
over time (> 4 weeks)
 When tested against dosing simplification,
educational interventions less robust
 Collaborative care (systematic inclusion of multiple
providers) superior to education alone intervention
Van Dulmen S, et al. Patient adherence to medical treatment: A review of reviews, BMC Health
Services Research, 7:55, 2007
29
What Works? A Review of Reviews

Social Support Interventions (practical, emotional,
undifferentiated)
 Large effect sizes seen with social support in welldesigned studies
Van Dulmen S, et al. Patient adherence to medical treatment: A review of reviews, BMC Health
Services Research, 7:55, 2007
30
What Works? A Review of Reviews

Complex or Multi-faceted Interventions (combine multiple
approaches)
 Less than half resulted in improved adherence, and
only a third better treatment outcomes
 Successful interventions very resource intensive
 Even the most effective did not yield large effect sizes
 Variability in intervention and study design
compromises assessment
Van Dulmen S, et al. Patient adherence to medical treatment: A review of reviews, BMC Health
Services Research, 7:55, 2007
31
Intervention Effects:
Largest to Smallest









Reduced drug dose frequency
Financial rewards
Prompting devices
Adherence-enhancing packaging
Telephone calls
Personal counseling
Home visits
Reminder letters
Written education material
32
Pitfalls to Avoid







Starting with the intervention “concept”
Doing too little
Intervening too late
Preaching to the choir
Not from a trusted source
Measurement via self-report
Broad intervention population
33
Recommendations for Improving
Adherence to Chronic Medications

Low hanging fruit
 Quick follow-up by medical staff after initial
prescription (not automated calls)
 Only apply sampling for cost reasons
 Get the discontinuers back!
34
Recommendations for Improving
Adherence to Chronic Medications

Medium term
 Screen for depression
 Build IT capacity to support clinical staff

Long term
 Targeted populations
 Medical Home
 Social support
35
Conclusions





Non-compliance remains an on-going and significant
problem in health care
The factors associated with non-compliance are now
being investigated
Literature reviews indicate that largest effect sizes will be
produced by complex or multi-faceted interventions
Multiple longitudinal interventions may be required to
obtain positive results
Non-compliance can significantly affect attainment of
P4P goals
36